Does Future Orientation Moderate the Relationship Between Impulse Control and Offending? Insights From a Sample of Serious Young Offenders

Abstract

Researchers have recently begun to examine motivational factors as moderators of the relationship between self-control and offending behavior. The current study extends prior work by investigating whether three aspects of future orientation (aspirations, expectations, and the use of future-oriented cognitive and behavioral strategies) play such a role. Drawing on 7 years of data from the Pathways to Desistance study (N = 1,333), we use hybrid effects negative binomial regression models to assess how within-individual changes in future orientation and impulse control are independently and jointly related to the offending variety of serious young offenders. Although impulse control and three components of future orientation had significant main effects on offending, no interaction between these components emerged in our results. Implications for future research are discussed.

Keywords

impulse control self-control future orientation young offenders

Gottfredson and Hirschi’s (1990) general theory of crime posits that the key individual difference that accounts for variation in criminal behavior is self-control, which they define broadly as “the tendency to avoid acts whose long-term costs exceed their momentary advantage” (Hirschi & Gottfredson, 1994, p. 3). In short, individuals with low self-control “will tend to be impulsive, insensitive, physical, risk-seeking, short-sighted, and nonverbal” (Gottfredson & Hirschi, 1990, p. 90). Previous research has consistently shown that self-control is a modest predictor of criminal involvement (Pratt & Cullen, 2000; Vazsonyi, Mikuška, & Kelley, 2017) and, as suggested by Gottfredson and Hirschi (1990), a range of analogous behaviors such as substance abuse (Tangney, Baumeister, & Boone, 2004), cheating and school outcomes (Muraven, Pogarksy, & Schmueli, 2006; Tangney et al., 2004), and poor interpersonal relationships (Evans, Cullen, Burton, Dunaway, & Benson, 1997). Thus, individuals with low self-control tend to be involved in a wide range of deviant activity, regardless of legality.

That said, in their original statement of the relationship between self-control and criminal behavior, Gottfredson and Hirschi (1990; see also Tittle, Ward, & Grasmick, 2004) offered an important caveat that “lack of self-control does not require crime and can be counteracted by situational conditions or other properties of the individual” (p. 89). Indeed, Pratt and Cullen’s (2000) meta-analysis showed that there tends to be a significant interaction between self-control and situational opportunity. Of particular interest to the present study, however, is the “other properties of the individual” piece of Gottfredson and Hirschi’s (1990) caveat. Specifically, in the current article, our concern is the extent to which aspirations, expectations, and the use of other future-oriented cognitive and behavioral strategies (FOCABS) moderate the relationship between a key component of self-control, the ability to control one’s impulses, and the variety of criminal behavior that adolescents and young adults engage in. Only a handful of studies have examined future orientation as a moderator of this relationship thus far (Chen & Vazsonyi, 2011; Clinkinbeard, 2014; Mahler, Simmons, Frick, Steinberg, & Cauffman, 2017; Robbins & Bryan, 2004). Here, we extend this body of work by drawing on data from serious offenders in the Pathways to Desistance (Pathways) study and using a hybrid effects negative binomial regression framework, the combination of which allows us to examine within-individual changes in future orientation and impulse control, and whether these jointly influence criminal involvement over the life course.

Self-Control, Future Orientation, and Criminal Involvement

In their work on moderators of self-control, Tittle, Ward, and Grasmick (2004) argue that there is a difference between the capacity to control oneself and possessing the motivation to do so. A person might, for instance, experience a strong impulse to commit a deviant act but simultaneously want to keep out of trouble. Looking at a sample of 350 adults from the Oklahoma City Survey, Tittle et al.’s (2004) study tapped self-control desire by looking at self-pride, praise/respect from valued others, perceptions of guilt, morality, and the probability of being caught with respect to a variety of hypothetical offenses. Behavioral measures of self-control ability (Grasmick, Tittle, Bursik, & Arneklev, 1993) and the subjective self-control desire measure had close-to-equal main effects on involvement in criminal behavior in Tittle et al.’s (2004) models. Perhaps more important, there was a significant interaction between the two; for instance, ability had much less predictive power when desire was high, and vice versa. Cochran, Aleska, and Chamlin (2006) similarly examined the roles of self-control ability and desire in academic dishonesty among 448 college students. Self-control ability was tapped with a 31-item scale similar to the Grasmick Scale, while questions about desire looked at the perceptions of getting caught, potential loss of respect of others, personal shame/guilt, moral condemnation of academic dishonesty, and individuals’ social maturity and integrity. The results of Cochran et al.’s (2006) analyses matched those of Tittle et al.’s (2004), with significant main effects for both ability and desire and a significant interaction between them.

Perhaps the principal overarching argument of Gottfredson and Hirschi (1990) is that individuals with low self-control are impulsive and tend to be stuck in the “here and now.” However, the work of Tittle and colleagues (2004) and Cochran et al. (2006) indicates that thinking about the potential consequences of criminal behavior—whether they be damage to one’s relationships, pride, or liberty—moderates the exercise of self-control. This suggests more broadly that being impulsive and thinking about one’s future are distinct phenomena within persons, a notion that has been explored by psychologists studying human development, personality, and psychopathology. For example, Zimbardo and colleagues’ work on time perspective theory has shown that items measuring present-hedonistic (i.e., characterized by impulsiveness, pursuit of present pleasures) and future-focused (i.e., characterized by ambition, future goals, planning) time orientations load onto separate, moderately correlated factors (Zimbardo & Boyd, 1999) and are differentially associated with outcomes such as substance abuse, risky driving, exercise, and sexual behavior (e.g., Henson, Carey, Carey, & Maisto, 2006; Keough, Zimbardo, & Boyd, 1999).

Other researchers have further distinguished between impulse control and future orientation at an empirical level. In a study examining the effect of these two constructs on delay discounting task performance, Steinberg and colleagues (2009) operationalized impulsivity using items capturing the propensity to act on the spur of the moment (i.e., motor impulsivity), an inability to delay gratification, and lack of perseverance, while the future orientation measure included items capturing one’s tendencies to plan ahead, to anticipate the future consequences of an action, and the relative prominence of a future, past, or present time perspective. Their analyses showed that the two constructs were modestly correlated (r = −.33; see also Copping, Campbell, & Muncer, 2014). More importantly, they found that although individual differences in one’s ability to weigh immediate versus future rewards were explained by differences in both impulse control and future orientation, age differences in this ability were explained only by age-related increases in future orientation. In light of this prior work, then, although it is likely that individuals with low future orientation will also score low on measures of impulse/self-control, it is not a given that all individuals will conform to that expectation (Chen & Vazsonyi, 2011; Clinkinbeard, 2014; Robbins & Bryan, 2004).

At a conceptual level, the construct of future orientation used in this prior research broadly refers to the “subjective construction of one’s future” which “provides the grounds for setting goals and making plans, exploring options and making commitments that consequently guide the person’s developmental course” (Seginer & Noyman, 2005, p. 18). As highlighted above, future orientation has been conceptualized in a variety of ways over the years (see Seginer, 2009; Steinberg et al., 2009); however, the tripartite model developed by Seginer and colleagues (Seginer, 2009; Seginer, Vermulst, & Shoyer, 2004) is perhaps the most well-developed and broadly encompassing. First, the motivational component of future orientation assumes that all behavior is goal-directed and that the motivation to pursue such goals is a function of their subjective value, the expectancy of success, and an individual’s efficacy beliefs (Bandura, 1991; Eccles & Wigfield, 1995). For instance, if an individual has high aspirations to attend a university, believes that he or she has the ability to do well in high school coursework, and that doing so will result in an offer of admission, the distal goal of attending college will motivate his or her proximal behavior to achieve that goal (Miller & Brickman, 2004). Second, as Seginer (2009) notes, the cognitive component of future orientation reflects the frequency of thought about prospective life-course domains such as education, work, and family, as well as existential concerns about oneself and others. There is also an affective aspect to future-oriented cognition, wherein individuals ascribe positive (hopes) and negative (fears) valence to their thinking about life-course domains and existential concerns (Markus & Nurius, 1986). Finally, the behavioral component of future orientation involves putting motivation and cognition about the future into concrete action (Seginer et al., 2004). This involves both exploration of future options (e.g., assessing opportunities, seeking advice about options, measuring options against one’s goals) and the commitment to pursue a course of behavior (Seginer, 2009).

Scholars have previously hypothesized about and empirically tested the roles of these various components of future orientation in antisocial behavior. Expectancy-value theory (Eccles & Wigfield, 1995), for instance, suggests both aspirations for future goals and expectations to achieve those goals play a role in whether an individual engages in future-oriented behavior (e.g., avoiding criminal activity, doing homework instead). An individual who aspires to have a good career may choose to avoid situations that could threaten that goal, while a person with low expectations about achieving a good life might feel there is nothing to lose by seeking shorter term gratifications. In that regard, Mahler, Simmons, Frick, Steinberg, and Cauffman (2017) note that some scholars have suggested expectations might be the more important of the two, as these “represent more concrete, realistic future beliefs that serve as the necessary link between aspirations and actual achievement” (p. 1504; see also Reynolds & Pemberton, 2001). Mahler and colleagues’ (2017) own research indicated, to the contrary, that levels of both aspirations and expectations have significant main effects of self-reported offending (SRO). However, other work by Knight, Ellis, Roark, Henry, and Huizinga (2017) has shown that while aspirations and expectations are associated with contemporaneous levels of offending, only expectations are associated with such behavior a year later.

Since the cognitive and behavioral aspects of future orientation represent the more concrete strategies individuals use to put aspirations into action, we would similarly hypothesize higher scores on measures of these components to be associated with lower levels of offending. In an early study assessing such relationships, Trommsdorff and Lamm (1980) found that, when compared with nondelinquent youth, delinquents are less optimistic about their futures, do not think as far into the future, and spend less time doing things that are future-oriented (e.g., thinking and talking about the future). Likewise, in a study of previously adjudicated adolescents, Robbins and Bryan (2004) found that better attitudes about the future (e.g., “what happens to me in the future depends mostly on me”) are associated with lower levels of risk behaviors such as drug and alcohol use and unsafe sex.

Although the main effects of future orientation’s components on antisocial behavior are important, their potential moderating effects on the relationship between impulse control and offending are of most interest to the current study. Drawing from self-discrepancy theory (e.g., Markus & Nurius, 1986), Silver and Ulmer (2012) have posited that the contemplation of possible selves might motivate the exercise of self-control; that is, when one perceives that an action may prevent the realization of a desired future self or make a feared self more likely, he or she might feel compelled to choose an alternative course of action in order to mitigate feelings of anxiety or guilty (Carver, Lawrence, & Scheier, 1999). Thus, as Tittle and colleagues (2004; see also Cochran, Aleska, & Chamlin, 2006) suggest, although some people “may have a strong capacity for self-control but may not always want to exercise it, while others may have weak self-control ability but have such a keen interest in controlling their deviant impulses that they end up conforming” (p. 147). It might be possible, for example, that a graduate student has difficulty controlling his impulses while out with friends on the weekend and often wakes up with regrets and a headache to show for it. Recognizing that such behavior might threaten his relationships, reputation, and ability to think and write clearly, the student might choose to avoid bars altogether or to limit nights out to once a month. In this example, the effect of the graduate student’s poor impulse control on deviant activity is mitigated by his orientation toward the future.

To date, only a handful of studies have probed whether such a moderated relationship exists among offenders. In terms of cross-sectional research, Robbins and Bryan (2004) examined the effect of future orientation on the relationship between impulsivity and a range of risk behaviors (e.g., alcohol and drug use, risky sexual behavior) within a sample of adjudicated adolescents. As noted above, both impulsivity and future orientation had significant main effects on these risk behaviors. For the alcohol use outcome, a significant interaction term indicated a stronger negative relationship with future orientation when levels of impulsivity were higher. Clinkinbeard (2014) reached similar conclusions in her analysis of baseline data from the National Longitudinal Study of Adolescent to Adult Health (Add Health). Low self-control and education (e.g., likelihood of attending college) and life (e.g., likelihood of living past age 35) expectations each had significant main effects on the frequency of criminal behavior. Moderation analyses further indicated that self-control had less of an effect on offending when achievement expectations were high.

Longitudinal studies on the moderating effect of future orientation have been conducted by Chen and Vazsonyi (2011) and Mahler and colleagues (2017). Chen and Vazsonyi (2011) conducted growth curve analyses using Waves 1–3 of the Add Health data. Results showed that Wave 1 levels of future expectations in the life and education domains, as well as impulsivity, were correlated with Wave 1 levels of antisocial behavior and predictive of rates of change through the latter two waves. Further, future expectations in the life domain moderated the relationship between Wave 1 impulsivity and levels of antisocial behavior across time. Mahler et al. (2017) examined data on low-level young offenders sampled during the Crossroads study. Their analyses again showed that an interaction between future aspirations and impulse control was a significant predictor of SRO 1 year later.

Considering these studies, there is preliminary evidence to suggest that future orientation serves to motivate individuals to resist impulses and avoid antisocial behavior in service of longer term goals. This prior work has, however, been limited in several ways. First, only two of these studies capture these relationships over time and, even so, cover a quite limited range (i.e., three follow-up periods each). This is potentially problematic given that prior research has reported substantial intraindividual variation in components of self-control and future orientation over the life course (Burt, Sweeten, & Simons, 2014; Monahan, King, Shulman, Cauffman, & Chassin, 2015; Steinberg et al., 2009). Second, measures of future orientation used in these studies have been narrow in scope. Those using the Add Health data (i.e., Chen & Vazsonyi, 2011; Clinkinbeard, 2014) relied on items that primarily capture the expectancies component of future orientation, and as such, the examination of FOCABS has not been possible. Conversely, in the Robbins and Bryan (2004) study, the Future Orientation Scale captures the cognitive and behavioral components but not the motivational aspect of future orientation. Further, while they examine a range of antisocial behaviors as outcome variables, serious criminal behaviors are not included.

The Current Study

The aim of the current study was to assess the relationships between impulse control, future orientation, and criminal offending. Although a complete measure of self-control (e.g., the Grasmick Scale) was not available in our data, focusing primarily on impulse control is appropriate for two reasons. First, several studies have shown that impulsivity has the most explanatory power of the six self-control dimensions (see Arneklev, Grasmick, & Bursik, 1999; Arneklev, Grasmick, Tittle, & Bursik, 1993; Longshore, Turner-Rand, & Stein, 1999). Second, as Tittle et al. (2004) have observed, most of Gottfredson and Hirschi’s (1990) key statements about the quality or trait of low self-control refer to an incapability to resist impulses.

Based on the research reviewed above, our hypotheses were as follows. First, in line with prior literature examining impulse control and self-control more broadly (e.g., Moffitt et al., 2011; Vazsonyi et al., 2017), we hypothesized that within-individual increases in impulse control would correspond with decreases in participants’ variety of criminal offending. Second, corresponding with previous literature on the links between future orientation and offending (e.g., Knight et al., 2017; Mahler et al., 2017; Trommsdorff & Lamm, 1980), we also hypothesized that increases in each of the components of future orientation (i.e., aspirations, expectations, and FOCABS) would be associated with reductions in offending. Finally, given the arguments of Tittle and colleagues (2004) and Silver and Ulmer (2012) about the importance of considering motivators of self-control and the findings of subsequent research examining such links (e.g., Chen & Vazsonyi, 2011; Clinkinbeard, 2014), we hypothesized that the strength of the relationship between impulse control and offending variety would be increasingly diminished as scores on measures of future orientation increased.

Although we aimed to test similar hypotheses to those assessed in prior research on the topic, the present study builds upon this work in several important ways. We utilize 10 waves of data on serious adolescent offenders collected during the Pathways study. These data allowed for the analysis of our key relationships across time and, by using hybrid effects regression techniques (Allison, 2009), to examine the effects of within-individual change in levels of key predictors on offending. We adopt such a within-individual analytical approach for several reasons. At the empirical level, some prior research has shown impulse/self-control to be stable in either absolute or relative terms throughout adolescence (e.g., Ray, Jones, Loughran, & Jennings, 2013; Turner & Piqeuro, 2002); however, other studies observe levels of impulse/self-control to increase throughout this developmental period (Shulman, Harden, Chein, & Steinberg, 2015), and that individuals’ rates of change can vary quite substantially from one another (Burt et al., 2014; Harden & Tucker-Drob, 2011). Recent work with the Pathways data has also shown that mean growth trajectories of impulse control and future orientation are approximately linear between ages 14 and 26, although growth of the latter tapers off in young adulthood (see Dmitrieva, Monahan, Cauffman, & Steinberg, 2012), but also that different groups of offenders (e.g., persistent vs. low rate) exhibit variations from these average trajectories (Monahan, Steinberg, Cauffman, & Mulvey, 2009). Theoretically, too, growth in future orientation during this period is expected, given that adolescents are under increasing social pressure to conform to age-normative developmental tasks such as the completion of school, finding employment, and building a family (Nurmi, Poole, & Kalakoski, 1994).

Given that the Pathways study is wholly comprised of serious offenders (Mulvey, Schubert, & Piquero, 2014), these data also are arguably more appropriate for studying factors underlying changes in offending than nationally representative data sets. As Cusick and Courtney (2007) show, for example, within the Add Health data used previously to address our research questions (Chen & Vazsonyi, 2011; Clinkinbeard, 2014), SRO variety scores over the previous 12 months tended to be quite low. Further, Mears, Cochran, and Beaver (2013) have shown that very few participants in the Add Health study score at the bottom end of the self-control measure. Sullivan and Loughran (2014) argue that it is precisely this lower end of the distribution that needs to be heavily populated when examining the form of the relationship between self-control and offending. The Pathways data fulfill this requirement more readily and, accordingly, allow us to assess the determinants of change in criminal behavior among the small proportion of serious offenders most concerning to the criminal justice system, which is an important aspect of studying change and stability in offending behavior (Laub & Sampson, 2001).

Method

Sample

Data were drawn from the Pathways study, an examination of serious offenders from adolescence through young adulthood. From November 2000 to January 2003, the original Pathways researchers used court records to identify and approach young offenders charged with a felony or serious misdemeanor (i.e., weapons offense, sexual assault) in Philadelphia County, PA, and Maricopa County, AZ. Of the 2,008 approached for inclusion in the study, 1,354 agreed to participate (Philadelphia n = 700, Maricopa n = 654; Mulvey et al., 2014). These two counties were chosen as research sites based on the presence of young offender populations with high rates of serious offending, the racial/ethnic diversity of these populations, and large enough female offender populations to allow for the examination of gendered differences (Mulvey et al., 2014). At baseline, the sample had a mean age of 16.04 years (standard deviation [SD] = 1.14), was primarily male (86%), and had a sufficient racial/ethnic mix (41.43% Black, 33.53% Hispanic, 20.24% White, and 4.80% Other). Participants completed a structured, computer-assisted interview 11 times between March of 2003 and March of 2010 (i.e., baseline, 6, 12, 18, 24, 30, 36, 48, 60, 72, and 84 months) to collect information on background characteristics, indicators of individual functioning, psychosocial development and attitudes, family context, community context, and personal relationships (Schubert et al., 2004). We draw on 10 waves for analyses in the current study, with the baseline data omitted, given that one of our control variables (i.e., street time) was not collected at the initial interview.

Measures

Dependent variable

The dependent variable of interest in this study is a measure of participants’ variety of offending behavior. This information was drawn from the SRO Scale, an adaptation of an earlier instrument developed by Huizinga, Esbensen, and Weiher (1991). The SRO Scale asks participants to report on their offending behavior during each recall period for a list of 22 offenses (e.g., number of times sold marijuana). The variety score used in the current analyses represents a count of the items endorsed, with higher scores indicating greater variety of offending. We opt for the variety score rather than frequencies, given that the former tends to be more reliable and less prone to being dominated by nonserious crime types that tend to occur with high frequency (e.g., minor property crimes, drug offenses; see Sweeten, 2012). Note that we provide (a) a set of zero-order correlations between the dependent variable, key independent variables, and continuous controls in Appendix Table A1 and (b) the items used to operationalize our key independent variables in Appendix Table B1.

Key independent variables

Impulse control

The ability to control one’s reactions to impulses was tapped using the Impulse Control subscale of the Weinberger Adjustment Inventory (IC-WAI; Weinberger & Schwartz, 1990). The subscale consists of 8 items (e.g., “I say the first thing that comes into my mind without thinking about it”) scored on a 5-point Likert-type scale (1 = false, 5 = true).² Seven of the eight items are reverse-coded, and the mean score is used for our analyses, with higher scores indicating greater control over impulses. Looking at the baseline wave of the Pathways study, Monahan, Steinberg, Cauffman, and Mulvey (2009) have previously shown this scale to have adequate reliability and a good fit to the data (α = .78, NFI = .95, CFI = .95, RMSEA = .07[Please insert expansion for NFI, CFI, and RMSEA).

Future-oriented motivation

We capture the value and expectancy components of future-oriented motivation using, respectively, the Aspirations and Expectations subscales of the Perceptions of Chances for Success measure, which was adapted from the work of Menard and Elliott (1996) for the Pathways study. Participants are asked seven pairs of questions designed to tap into their aspirations (e.g., “How important is it to you to have a good job or career?”) and expectations (e.g., “What do you think your chances are to have a good job or career?”) about work, family, and law-abidance in the future. Likert-type scale response options ranged from 1 (not at all important/poor) to 5 (very important/excellent). Mean scores for both scales are used in our analyses, with higher scores indicating greater aspirations or expectations for the future. Prior analyses have shown adequate reliability for both the Aspirations (α = .67) and Expectations (α = .81) Scales (Pathways to Desistance, n.d.).

FOCABS

This study operationalizes the cognitive and behavioral elements of future orientation using the Future Outlook Inventory (FOI), developed using items from the Life Orientation Task (Scheier & Carver, 1985), the Zimbardo Time Perspective Inventory (Zimbardo, 1990), and the Consideration of Future Consequences Scale (Strathman, Gleicher, Boninger, & Edwards, 1994). Participants are asked to rate, on a 4-point Likert-type scale, how true a set of 8 items are to themselves (1 = never true, 4 = always true). These items tap the extent to which the individual thinks about the future (e.g., “I think about how things might be in the future”) and the types of behaviors they use to plan and prepare for the future (e.g., “I will keep working at difficult, boring tasks if I know they will help me get ahead later”). We use the mean of the 8 items, with higher scores indicating more use of FOCABS. Prior work by Monahan et al. (2009) has shown that the FOI has an excellent fit to the baseline Pathways data and adequate reliability (α = .68, NFI = .96, CFI = .97, RMSEA = .03).

Control variables

Several time-variant and time-invariant covariates are controlled for in the analyses for this study. At the social/environmental level, we control for three of these. First, given that previous research indicates witnessing or being a victim of violence is associated with depressed development of future orientation and impulse control (Monahan et al., 2015), the sum total of items endorsed on the Exposure to Violence Inventory is entered as a control in our analyses. Second, both future orientation (Kerpelman, Eryigit, & Stephens, 2008) and criminal behavior (Colvin, Cullen, & Vander Ven, 2002) have also been linked to perceived social support. To account for this, the Depth of Social Support subscale from the Contact with Caring Adults Inventory is included. This measure is a count of the number of adults mentioned by participants as sources of support in three or more of the eight domains covered by the inventory (e.g., “Adults with whom you can talk about important decisions”). Finally, previous research has shown that social learning via peers impacts offending (Pratt et al., 2010) and future orientation (Brickman & Miller, 2001). Thus, we control for the impact of antisocial peers on our key variables of interest using the mean rating on the Peer Antisocial Behavior subscale of the Peer Delinquent Behavior measure. Participants rated the prevalence of antisocial behavior among their friends for 12 types of behavior on a 5-point Likert-type scale (1 = none of them, 5 = all of them).

Several variables are included to control for individual-level confounding as well. First, substance use has previously been associated with criminal behavior, impulsivity, and future orientation (Robbins & Bryan, 2004). As such, we control for this using the mean score of dependency symptoms subscale of the Substance Use/Abuse Inventory. Participants were asked a series of 10 questions (e.g., “Have you ever wanted a drink or drugs so badly that you could not think about anything else?”) on this measure. Higher scores signify greater dependence on drugs and/or alcohol. Second, the ability to regulate one’s emotions has also been linked to several forms of offending behavior (Roberton, Daffern, & Bucks, 2012), impulsivity (Davidson, Putnam, & Larson, 2000), and future orientation (Marroquin, Boyle, Nolen-Hoeksema, & Stanton, 2016). Thus, the mean score from the 12-item Children’s Emotion Regulation Scale is included to account for its influence on the key relationships of interest. Higher scores on this measure indicate greater control over one’s emotions. Third, we account for participants’ ages given its connection to future orientation, impulse control (Steinberg et al., 2009), and criminal offending (Sweeten, Piquero, & Steinberg, 2013). Fourth, although gendered differences in future orientation are not found consistently (Steinberg et al., 2009), we include a dummy-coded variable (0 = male, 1 = female) given the more consistent findings on gender differences in self-control and criminal behavior (LaGrange & Silverman, 1999). Fifth, racial/ethnic group differences are controlled for using a series of dummy variables, with Whites being the reference category against which Blacks, Hispanics, and individuals in the Other race category are compared. Finally, given that many participants spent time in correctional facilities during the study period and that this reduced their opportunities to offend (Piquero et al., 2001), we include a measure of the proportion of time during each recall period that participants spent in secure correctional settings without community access. Higher scores on this measure mean that participants had less time in the community to offend.

Analytical Strategy

To model the relationships of impulse control and future orientation with offending, we employed a series of hybrid effects negative binomial regression models (Allison, 2009). The assumption of a negative binomial distribution is appropriate to the data given the overdispersion of participants’ offending variety (Osgood, 2000). Specifically, there was a significant amount of variation relative to the mean (M = 1.34, SD = 2.35), and the distribution was quite skewed given the large proportion of participants who had not offended at each recall period (i.e., zeros made up 49.17% of the distribution). To test our hypotheses and model the dependency of observations given the repeated measures, the hybrid effects regression technique is advantageous over strict random or fixed effects models. Unlike random effects models, fixed effects equations do not assume that unmeasured variables are uncorrelated with the error term (Firebaugh, Warner, & Massoglia, 2013). With repeated measures, these equations are able to remove the effect of time-invariant confounders, whether measured or not (Allison, 2009), although there is still potential for omitted variable bias due to unmeasured time-variant confounding (Bjerk, 2009; Firebaugh, Warner, & Massoglia, 2013). However, a resultant drawback of the fixed effects approach is the inability to model the impact of measured time-invariant variables (e.g., gender). Further, these equations do not allow for the estimation of between-person effects.

The hybrid effects approach allows us to overcome both of these drawbacks while retaining the ability to control for unmeasured time-invariant confounders. To do so, we partition time-varying covariates into person-specific means for the duration of the study (time-stable, between-individual component) and individual deviation scores from these means for each recall period (time-variant, within-individual component; see Allison, 2009; Firebaugh et al., 2013). These pairs of components are entered into a random effects negative binomial regression equation, along with other time-stable control variables. Two such models are presented in the present article. Model 1 presents only the independent effects of the variables of interest, and Model 2 includes interaction terms to capture the potential moderating effects of the time-variant future orientation constructs on the impulse control-offending relationship. All models included person-specific measures of the number of months covered in the recall period in order to account for the fact that the times between interviews were not equal for all participants in each wave of the study.

Results

The demographic characteristics and descriptive statistics for the time-stable and time-variant components of each variable are reported in Table 1. Over the course of the study, the average offending variety score was quite low (M = 1.34, SD = 2.35); however, there was substantial variation from this mean value, with scores ranging from 0 to 19. At the first wave of data included in the current study, participants’ ages ranged from 14 to 20 years of age (M = 16.55, SD = 1.15). The majority of the sample was male (85.67%). Approximately, 40% identified as being African American, 34% were Hispanic, 21% were White, and approximately 5% identified as being some other race or ethnicity.

Table 1.

Descriptive Statistics.

Variable	%/Mean (SD)	Range
Offending variety	1.34 (2.35)	0–19
Ethnicity
White	21.04	—
African American	40.03	—
Hispanic	34.21	—
Other race	4.72	—
Gender (male)	85.67	—
Age (at baseline)	16.55 (1.15)	14–20
Impulse control
Time stable	3.22 (0.76)	1.08–4.98
Time variant	0.00 (0.60)	−3.05–2.82
FOCABS
Time stable	2.62 (0.39)	1.45–3.83
Time variant	0.00 (0.40)	−1.99–1.80
Aspirations
Time stable	4.48 (0.38)	2.93–5.00
Time variant	0.00 (0.41)	−3.03–1.66
Expectations
Time stable	3.66 (0.66)	1.52–5.00
Time variant	0.00 (.60)	−2.86–2.53
Emotion regulation
Time stable	2.82 (0.48)	1.29–4.00
Time variant	0.00 (0.49)	−2.27–2.00
Substance use
Time stable	0.60 (0.99)	0.00–7.80
Time variant	0.00 (1.35)	−7.80–9.00
Peer ASB
Time stable	1.73 (0.51)	1.00–3.77
Time variant	0.00 (0.60)	−2.59–3.42
Exposure to violence
Time stable	1.15 (1.05)	0.00–6.80
Time variant	0.00 (1.40)	−6.80–8.60
Support depth
Time stable	0.81 (0.29)	0.00–2.00
Time variant	0.00 (0.46)	−1.40–1.70
Street time
Time stable	0.33 (0.29)	0.00–1.00
Time variant	0.00 (0.29)	−0.90–0.89

Note. ASB = antisocial behavior; FOCABS = future-oriented cognitive and behavioral strategies.

The time-stable components in Table 1 represent the participants’ mean score on each measure for the ten waves of data included in the study. The mean aspirations score, for example, was 4.48 (SD = 0.38) and ranged from 2.93 to 5.00. This suggests that participants’ aspirations for careers, family, and adherence to the law in the future were quite high. On the other hand, the mean expectations score was 3.66 (SD = 0.66), with a range from 1.52 to 5.00, indicating that study participants had, on average, lower future expectations about work, family, and law-abidance. The means of the time-variant components do not reveal much on their own; however, examining their ranges, as displayed in Table 1, shows that participants exhibited some change on these variables throughout the study period. For instance, the mean deviation score for impulse control was zero (SD = 0.60), but there was an observed range of −3.05–2.82. Thus, participants deviated both positively and negatively, and in some cases quite substantially, from their mean levels of impulse control. Our other key variables (i.e., aspirations, expectations, FOCABS) similarly vary quite widely at the within-individual level.

Table 2 presents the results of hybrid effects negative binomial regression models with (Model 1) and without (Model 2) interactions between impulse control and measures of future orientation. Both models have an effective sample size of 11,864 observations, reflecting data collected from 1,333 individuals over a mean of 8.9 recall periods. We report bootstrapped standard errors (SEs) based on 50 replications for all coefficient estimates, as suggested by Allison (2009) and Cameron and Trivedi (2005). Note that this analytic sample is slightly smaller than the full sample (N = 1,354), due primarily to item nonresponse. To check whether our analytical sample differed significantly from the excluded sample, we ran a series of t tests comparing the groups’ means on the variables included in our models at each time interval. Overall, the groups were very similar, with only 11 of the 130 tests indicating statistically significant differences in group means. A table reporting the significant between-group differences can be found in Appendix Table C1.

Table 2.

Hybrid Effects Negative Binomial Regression Results.

	Model 1: Without Interaction Terms			Model 2: With Interaction Terms
	b	SE ^a	Exp(b)	b	SE ^a	Exp(b)
Time-variant factors
Impulse control	−0.191***	.021	0.826	−0.192***	.019	0.825
FOCABS	−0.088*	.037	0.916	−0.086**	.029	0.918
Aspirations	−0.013	.031	0.987	−0.011	.032	0.989
Expectations	−0.124***	.019	0.884	−0.128***	.021	0.880
Emotion regulation	−0.048	.026	0.953	−0.047	.029	0.954
Substance use	0.084***	.008	1.088	0.084***	.007	1.087
Peer ASB	0.291***	.022	1.338	0.291***	.017	1.338
Exposure to violence	0.138***	.008	1.148	0.138***	.007	1.148
Support depth	−0.010	.024	0.990	−0.009	.025	0.991
Street time	−0.150**	.045	0.861	−0.150***	.032	0.860
Age	−0.169***	.007	0.845	−0.169***	.007	0.845
Time-stable factors
Male (reference)
Female	−0.390***	.066	0.677	−0.390***	.070	0.677
White (reference)
African American	−0.216***	.053	0.806	−0.215***	.053	0.806
Hispanic	−0.129*	.055	0.879	−0.128*	.055	0.880
Other race/ethnicity	−0.233*	.118	0.792	−0.233*	.115	0.792
Interaction terms
Impulse control × FOCABS				0.034	.044	1.035
Impulse control × aspirations				0.018	.062	1.019
Impulse control × expectations				−0.037	.036	0.964
Constant	1.385***	.336	3.995	1.382***	.334	3.983
Wald $2 χ^{2}$	11,024.64***			9,032.28***
Log likelihood	−14,918.686			−14,917.823

Note. N = 1,333; NT = 11,864; SE = standard error; ASB = antisocial behavior; FOCABS = future-oriented cognitive and behavioral strategies.

^a Based on 50 bootstrap replications.

*p < .05. **p < .01. ***p < .001.

Turning to Model 1, three of our four key time-variant variables were statistically significant (time-invariant components reported in Appendix Table D1). Specifically, and as expected, within-individual changes in impulse control, FOCABS, and expectations for the future had significant negative associations with offending variety. Although Table 2 reports the unstandardized regression coefficients and bootstrapped SEs for these relationships, their substantive significance is best captured by the exponentiated coefficients (i.e., Exp[b]). These coefficients can be interpreted as the percent change in the expected count of crime types participants report in a recall period for each unit increase in the independent variable. Coefficients less than 1.00 represent the percent decrease, while those greater than 1.00 represent the percent increase. For example, when holding all else constant, the exponentiated coefficient for impulse control of 0.826 in Model 1 suggests that a single-unit increase from one’s mean level was associated, on average, with an expected decrease of 17.4% in offending variety. Likewise, one-unit increases in levels of FOCABS and future expectations led to expected decreases in offending variety of 8.4% and 11.6%, respectively. Contrary to hypotheses, change in aspirations had no statistically significant association with our dependent variable.

For the remaining time-variant control variables in the model, within-person changes in substance use, peer antisocial behavior, and exposure to violence were associated with increases in offending variety of 8.8%, 33.8%, and 14.8%, respectively, all of which were significant at the p < .001 level. Conversely, and as expected, increases in the amount of time spent in secure settings during a recall period and age were associated with reductions in the numbers of unique crime types engaged in during a recall period. Within-individual changes in emotion regulation and depth of social support were not significantly related to participants’ variety of criminal offending. In terms of the time-stable controls, females’ offending variety scores were approximately 32.3% less than those of males, on average. All three dummy variables for racial/ethnic backgrounds were statistically significant as well. Specifically, when compared with Whites, the average offending variety of African American participants was approximately 19.4% less, that of Hispanic was 12.1% less, and 20.8% less for those of another race/ethnicity.

For Model 2, the results of which are displayed in the right panel of Table 2, we included interaction terms between the time-variant components of impulse control and FOCABS, aspirations, and expectations. A likelihood ratio test indicated that including these terms did not result in a significant improvement in fit to the data over Model 1, χ² $2 X^{2}$ (3, N = 11,864) = 1.71, p > .05. The results for the time-variant and time-stable main effects remained quite similar to those in Model 1 in terms of the size of coefficients, statistical significance, and directions of influence. However, the terms representing interactions between impulse control and FOCABS (b = .034, SE = .044, p > .05), aspiration (b = .018, SE = .072, p > .05), and expectations (b = − .037, SE = .036, p > .05) did not emerge as being significantly associated with change in offending variety scores. For example, the interaction between impulse control and FOCABS is displayed in Figure 1. As the plot further illustrates, the slope for impulse control does not vary based on different levels of the FOCABS measure.

Figure 1.

Interaction between impulse control and future-oriented cognitive and behavioral strategies on variety of offending.

Since several previous studies have found that future orientation moderates the relationship between impulse/self-control and offending, sensitivity analyses were conducted in light of the null findings on this relationship in the current study. As a first step in this process, each of the three interaction terms was added to Model 1 separately. These analyses indicated that none of the three measures of future orientation had a statistically significant interaction with impulse control in their relationship to offending variety scores. Second, interaction terms were entered into the model both jointly and separately, with the future orientation measures lagged by one period. Interaction terms were not significant in any of these models. Third, we entered all time-varying covariates into the model and lagged by one period (save for age and street time). Again, none of the interaction terms were statistically significant. Finally, we constructed terms interacting the time-stable impulse control component with time-stable components for aspirations, expectations, and FOCABS. When these were entered into Model 1 both jointly and separately, none reached the level of statistical significance.

Discussion

Previous research has shown that the relationship between self-control and criminal behavior is moderated by “other properties of the individual” (Gottfredson & Hirschi, 1990, p. 89), such as pride, guilt, morality, and perceptions of respect from the individuals one is bonded to (e.g., Cochran et al., 2006; Tittle et al., 2004). More recently, Silver and Ulmer (2012) have hypothesized further that “individuals [also] evaluate courses of action…in terms of whether they are right or wrong…in the context of desired or undesired conceptions of themselves in the future” (p. 701). In other words, the degree to which one thinks about, plans for, and expects good things for the future may alter his or her exercise of self-control and, consequently, his or her behavior in the present. Empirical investigation of this hypothesis has been confined to a handful of studies (Chen & Vazsonyi, 2011; Clinkinbeard, 2014; Mahler et al., 2017; Robbins & Bryan, 2004) that, as discussed at the outset of this article, are limited either by the data used or the statistical procedures employed. Thus, the goal of the current study was to build upon prior work and evaluate whether several elements of future orientation moderate the relationship between impulse control and offending. To do so, we used hybrid effects negative binomial regression models (see Allison, 2009) to analyze the within-individual changes pertinent to this hypothesized relationship in a sample of serious adolescent and young adult offenders.

Looking first at independent effects, based on Gottfredson and Hirschi’s (1990) exposition of their theory as well as previous examinations of the individual components of self-control (e.g., Arneklev et al., 1993; Longshore et al., 1999), we hypothesized that change in impulse control would have a significant effect on participants’ criminal involvement. Indeed, results from both models presented here indicated that positive deviations from one’s average level of impulse control were associated with reductions in the variety of criminal behaviors he or she engaged in during a given recall period. This result provides further support for the notion that self-control and its constituent elements are robust predictors of whether one engages in crime (Pratt & Cullen, 2000; Vazsonyi et al., 2017). More encouraging, though, is that the current study also shows that there are meaningful within-individual changes in impulse control across the life course and that these have substantive impacts on antisocial behavior (see also Burt et al., 2014; Monahan et al., 2009). This finding, which falls in line with recent research on the development of impulse control during adolescence and young adulthood (e.g., Burt et al., 2014; Harden & Tucker-Drob, 2011; Shulman et al., 2015; although see Ray et al., 2013), is contrary to Gottfredson and Hirschi’s (1990) assertion that self-control (and its constituent components) will be relatively stable beyond age 10.

The second hypothesis of this study was that future orientation would also have a significant independent effect on the variety of participants’ offending. Specifically, drawing from previous work on expectancy-value theory (e.g., Eccles & Wigfield, 1995) and future orientation (see, especially, Seginer, 2009), we examined the impact of changes in the aspirations, expectations, and use of FOCABS by participants. The results of both models showed that for members of our sample, the variety of criminal behaviors individuals were involved in was reduced by intraindividual increases in (a) positive expectations about the future and (b) the degree to which participants thought and acted in a future-oriented manner (e.g., thinking about the consequences of one’s actions, making plans or to-do lists). On the other hand, changes in aspirations did not have a significant effect on offending. Taken together, these findings bolster support for previous studies that have shown expectancy/self-efficacy beliefs (Cuevas, Wolff, & Baglivio, 2017; Iselin, Mulvey, Loughran, Chung, & Schubert, 2012) and cognitive–behavioral (Robbins & Bryan, 2004) components of future orientation to be negatively associated with criminal behavior. They also suggest, in line with the arguments of Reynolds and Pemberton (2001), that expectations represent more concrete beliefs about putting aspirations into action and, accordingly, will have a larger influence on actual behavior. Further, our results indicate that it is important to consider not only offenders’ expectations for a good future as impacting their criminal involvement but also the degree to which they are able to translate such expectations into concrete cognitive and behavioral strategies in the present.

In terms of the focal hypothesis of this study, we expected that all three variables used to operationalize aspects of future orientation would moderate the relationship between impulse control and offending variety. Contrary to prior research (Chen & Vazsonyi, 2011; Clinkinbeard, 2014; Mahler et al., 2017; Robbins & Bryan, 2004), our analyses found no such associations. Although we can only speculate as to the reasons for these null findings, the most likely explanation lies in the divergent distributions of offending, impulse control, and future orientation between samples used in Pathways and those in previous studies. The sample drawn in Pathways comprises young offenders charged with a felony or misdemeanor weapons and sexual assault offenses (Schubert et al., 2004). Our own analyses of the baseline wave of these data showed that the mean age at first offense was 10.42 years (SD = 1.81), with 50.26% of participants admitting to committing their first offense at age 9 or earlier. By all accounts, then, the offending histories of these individuals would be classified as serious and indicative of early problems with self-regulatory capacity (Caspi, 2000; Moffitt et al., 2011).

The types of individuals who comprise the samples used in previous work on our research question are qualitatively different than those of Pathways. In the nationally representative Add Health data, used by Chen and Vazsonyi (2011) and Clinkinbeard (2014), the distributions for impulsivity and measures of future orientation make this clear. As Chen and Vazsonyi (2011) report, the distribution of impulsivity was approximately normal (skewness = .44), with a mean score of 2.27 and a standard deviation of .64. These values suggest that only ∼2.28% of the sample scored above 3.55 on a 5-point measure of impulsivity; that is, the Add Health comprises very few individuals who report being highly impulsive. Conversely, the descriptives for the 5-point impulse control measure in Pathways (M = 3.22, SD = 0.97, skewness = − .04) indicate a substantial proportion of participants populates this critical range. Univariate descriptives for 5-point scales capturing education (M = 4.49, SD = 0.80, skewness = −2.15) and life future orientation (M = 4.52, SD = 0.54, skewness = −1.27) in the Add Health data also indicate these distributions are negatively skewed and that the vast majority of participants reported quite high aspirations and expectations for the future. On the other hand, the 5-point scale for expectations (M = 3.65, SD = 0.89, skewness = − .23) and 4-point FOCABS Scale (M = 2.62, SD = 0.57, skewness = .12) from Pathways show that a larger group populates the low end of these distributions (i.e., participants are more likely to be less oriented to the future). Although the data used by Mahler et al. (2017) and Robbins and Bryan (2004) are explicit samples of the offending population, both note that the most common offenses were not particularly serious (e.g., vandalism, theft, drug possession), and an examination of the distributions of key variables yields similar results to those observed in the Add Health data.

The restricted variation in key measures for the data from previous studies may result in an identification problem for the estimation of the complete functional form of the relationship between impulse/self-control, future orientation, and criminal involvement (Sullivan & Loughran, 2014). In other words, because data such as the Add Health have limited numbers of participants populating the extreme ends of these distributions, identifying how differences in future orientation moderate the association between impulse control and offending (or vice versa) for individuals with serious deficits in these areas may be untenable. Sullivan and Loughran (2014) have explicated similar difficulties with estimating the nonlinearity of the relationship between self-control and offending with general population samples when compared to samples comprised of youth with more serious delinquent youth.

Although the study presented here adds to knowledge of the relationships between future orientation, impulse control, and criminal offending, it is not without its limitations. As noted earlier, the use of hybrid effects regression models in the present study allowed us to control for unobserved, time-invariant confounders that may have masked the potentially spurious nature of the relationships under investigation here (Firebaugh et al., 2013) and which, consequently, have been a long-standing source of methodological concern for developmental and life-course criminologists (e.g., Blokland & Nieuwbeerta, 2010; Nagin & Paternoster, 1991). However, fixed and hybrid effects models still rest on the assumption that all relevant time-variant predictors are included in the regression equation (Bjerk, 2009). In other words, it is still necessary to account for factors that may change over time and, in the case of our study, potentially confound the observed relationships between future orientation, impulse control, and individuals’ variety of offending. In the current study, we have controlled for several intraindividual as well as environmental factors that are subject to change over time and that are theoretically and/or empirically relevant to our research questions. Still, it is possible that we have failed to account for some time-variant construct(s) that would render our findings spurious.

Another set of limitations to this study relates to the measurement of some of the constructs included in our models. One of our key independent variables, a mean score from the FOI, contains items measuring both cognitive and behavioral elements of future orientation. While previous studies have shown that the cognitive and behavioral components of future orientation influence each other (e.g., Seginer & Noyman, 2005; Seginer et al., 2004), it is unclear in the results reported here precisely what relative influence each component has on impulse control and participants’ variety of offending. Future research, then, should endeavor to parse these effects using differentiated measurement.

Several items on the IC-WAI also resemble items we might find on a measure of sensation-seeking. For example, the item “I’m the kind of person who will try anything once, even if it’s not that safe” on the IC-WAI appears to be quite similar to an item used by Burt, Sweeten, and Simons (2014) to capture sensation-seeking: “You could do something most people would consider dangerous like driving a car.” Although previous work with the Pathways data by Monahan and colleagues (2009) suggests that the items used for the IC-WAI load onto a single factor that fits the data quite well (α = .78, NFI = .95, CFI = .95, RMSEA = .07), future research would do well to examine whether both impulse control and sensation-seeking have different relationships with future orientation in terms of a moderated impact upon antisocial behavior. This is an important question given that a mounting body of evidence suggests that the growth trajectories of impulse/self-control and sensation-seeking are quite different and that the two constructs have different effects on offending behavior (e.g., Burt et al., 2014; Harden & Tucker-Drob, 2011; Shulman et al., 2015).

We also recognize that there is potential conceptual overlap between a minority of items on the FOI (e.g., “I usually think about the consequences before I do something”) and those on the IC-WAI Scale (e.g., “I stop and think things through before I act”). This is to be expected, however, given both future orientation and impulse control fall under the conceptual umbrella of psychosocial maturity (Monahan et al., 2015; Steinberg & Cauffman, 1996). A close examination of the entirety of the items on these measures shows that while the IC-WAI is intended to capture the extent to which individuals are able to resist immediate gratification and control reactions to impulses, the FOI taps into their abilities to consider both proximal and distal concerns in decision-making and to plan for the future. The positive but still relatively weak bivariate correlation (r = .21, p < .001) between these two scales in our data further suggests that they are measuring empirically distinct constructs. Thus, despite some similarities when taken at face value, use of the two measures allows for greater precision in understanding how different elements of decision-making processes affect offending in adolescence and emerging adulthood.

Our use of a count measure of social support is perhaps not ideal. It would arguably have been more prudent to include variables that tap the degree of supervision, bonding, and support provided by participants’ parents, for example, given the empirical association between such factors and antisocial behavior (Pratt et al., 2010). However, in the Pathways data, such measures have substantial proportions of missing data, especially in the latter waves of the study, thus rendering their inclusion untenable. Given that parents and others to whom adolescents are closely bonded are a source of future-oriented beliefs and strategies (Kerpelman et al., 2008; Seginer et al., 2004), the inclusion of a wider range of variables tapping these constructs will be necessary in future studies. Finally, although others (e.g., Arneklev et al., 1993, 1999; Longshore et al., 1999) have shown impulse control to have the most predictive power of the aspects of self-control originally outlined by Gottfredson and Hirschi (1990), subsequent studies examining research questions similar to ours should aim to include a more encompassing measure of self-control as well. Changes in future orientation may, for instance, have an effect on the degree or type of risk-taking or sensation-seeking behaviors one engages in.

Conclusion and Implications

In the current study, we used data drawn from a sample of serious young offenders followed into adulthood to examine whether future orientation moderates the relationship between impulse control and criminal behavior. Hybrid effects negative binomial regression models indicated that there was no such interaction in this data set. Still both impulse control and future orientation had statistically and substantively significant relationships with offending variety. Despite potential limitations in measurement and model specification, this study advances knowledge about the connections between future orientation, impulse control, and offending in several meaningful ways. First, we have shown that within-individual change in impulse control and future orientation is relevant to changes in offending. Second, we have suggested that the moderating effects of future orientation may not apply to serious offenders who populate the extreme ends of self/impulse control, future orientation, and offending distributions that the data used in previous research do not capture well.

A third implication of our research dovetails from the previous two. Specifically, we have shown that intraindividual increases in future orientation are associated with decreased offending among serious offenders. Future orientation may thus be an appropriate responsivity factor or criminogenic need that could be targeted in the treatment of high-risk offenders. Researchers are increasingly directing attention toward pretreatment levels of motivation (e.g., Mossière & Serin, 2014; Ward, Day, Howells, & Birgden, 2004), as a large body of work suggests low motivation is a robust predictor of treatment program attrition and, consequently, subsequent reoffending (Olver, Stockdale, & Wormith, 2011). The fact that future orientation encompasses both motivation and cognitive–behavioral strategies needed to pursue goals (Eccles & Wigfield, 1995; Seginer & Noyman, 2005) suggests that the increasing attention being given to motivational interviewing (McMurran, 2009) and strategies to teach risk avoidance and cognitive restructuring within correctional programs (Landenberger & Lipsey, 2005) is apt to produce meaningful decreases in reoffending.

Footnotes

Appendix A

Appendix B

Appendix C

Appendix D

Acknowledgements

This research relies on data collected from the Pathways to Desistance study (ICPSR 29961).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Support and funding for the data collection were provided by the Arizona Governor’s Justice Commission (JBISA012244400), John D. and Catherine T. MacArthur Foundation, Pennsylvania Commission on Crime and Delinquency (2001-J05-011944, 2002-J04-13032, 2003-J04-14560, 2004-J04-15849, 2005-J04-17071, 2006-J04-18272), Robert Wood Johnson Foundation (043357), Centers for Disease Control and Prevention, National Institute on Drug Abuse (R01 DA 019697 05), National Institute of Justice (1999-IJ-CX-0053, 2008-IJ-CX-0023), Office of Juvenile Justice and Delinquency Prevention (2000-MU-MU-0007, 2005-JK-FX-K001, 2007-MU-FX-0002), William Penn Foundation, and William T. Grant Foundation (99-2009-099).

ORCID iD

Damon M. Petrich

References

Allison

(2009). Fixed effects regression models. Thousand Oaks, CA: SAGE.

Arneklev

B. J.

Grasmick

H. G.

Bursik

R. J.

(1999). Evaluating the dimensionality and invariance of “low self-control”. Journal of Quantitative Criminology, 15, 307–331.

Arneklev

B. J.

Grasmick

H. G.

Tittle

C. R.

Bursik

R. J.

Jr (1993). Low self-control and imprudent behavior. Journal of Quantitative Criminology, 9, 225–247.

Bandura

(1991). Self-regulation of motivation through anticipatory and self-reactive mechanisms. In Dienstbier

R. A.

(Ed.), Current theory and research in motivation, Vol. 38. Nebraska Symposium on Motivation, 1990: Perspectives on Motivation (pp. 69–164). Lincoln, NE: University of Nebraska Press.

Bjerk

(2009). How much can we trust causal interpretations of fixed-effects estimators in the context of criminality. Journal of Quantitative Criminology, 25, 391–417.

Blokland

A. A. J.

Nieuwbeerta

(2010). Considering criminal continuity: Testing for heterogeneity and state dependence in the association of past to future offending. The Australian and New Zealand Journal of Criminology, 43, 526–556.

Brickman

S. J.

Miller

R. B.

(2001). The impact of sociocultural knowledge on future goals and self-regulation. In McIrney

D. M.

Van Etten

(Eds.), Research on sociocultural influences on motivation and learning (pp. 119–137). Greenwich, CT: Information Age.

Burt

C. H.

Sweeten

Simons

R. L.

(2014). Self-control through emerging adulthood: Instability, multidimensionality, and criminological significance. Criminology, 52, 450–487.

Cameron

A. C.

Trivedi

P. K.

(2005). Microeconometrics: Methods and applications. Cambridge, UK: Cambridge University Press.

10.

Carver

C. S.

Lawrence

J. W.

Scheier

M. F.

(1999). Self-discrepancies and affect: Incorporating the role of feared selves. Personality and Social Psychology Bulletin, 25, 783–792.

11.

Caspi

(2000). The child is the father of the man: Personality continuities from childhood to adulthood. Journal of Personality and Social Psychology, 78, 158–172.

12.

Chen

Vazsonyi

A. T.

(2011). Future orientation, impulsivity, and problem behaviors: A longitudinal moderation model. Developmental Psychology, 47, 1633–1645.

13.

Clinkinbeard

S. S.

(2014). What lies ahead: An exploration of future orientation, self-control, and delinquency. Criminal Justice Review, 39, 19–36.

14.

Cochran

J. K.

Aleksa

Chamlin

M. B.

(2006). Self-restraint: A study on the capacity and desire for self-control. Western Criminology Review, 7, 27–40.

15.

Colvin

Cullen

F. T.

Vander Ven

(2002). Coercion, social support, and crime: An emerging theoretical consensus. Criminology, 40, 19–42.

16.

Copping

L. T.

Campbell

Muncer

(2014). Conceptualizing time preference: A life-history analysis. Evolutionary Psychology, 12, 829–847.

17.

Cuevas

Wolff

K. T.

Baglivio

M. T.

(2017). Self-efficacy, aspirations, and residential placement outcomes: Why belief in a prosocial self-matters. Journal of Criminal Justice, 52, 1–11.

18.

Cusick

G. R.

Courtney

M. E.

(2007). Offending during late adolescence: How do youth aging out of care compare with their peers? (Issue Brief No. 111). Chicago, IL: Chapin Hall Center for Children.

19.

Davidson

R. J.

Putnam

K. M.

Larson

C. L.

(2000). Dysfunction in the neural circuitry of emotion regulation: A possible prelude to violence. Science, 289, 591–594.

20.

Dmitrieva

Monahan

K. C.

Cauffman

Steinberg

(2012). Arrested development: The effects of incarceration on the development of psychosocial maturity. Development and Psychopathology, 24, 1073–1090.

21.

Eccles

J. S.

Wigfield

(1995). In the mind of the actor: The structure of adolescents’ achievement task values and expectancy-related beliefs. Personality and Social Psychology Bulletin, 21, 215–225.

22.

Evans

D. T.

Cullen

F. T.

Burton

V. S. J.

Dunaway

R. G.

Benson

M. L.

(1997). The social consequences of self-control: Testing the general theory of crime. Criminology, 35, 475–504.

23.

Firebaugh

Warner

Massoglia

(2013). Fixed effects, random effects, and hybrid models for causal analysis. In Morgan

S. L.

(Ed.), Handbook of causal analysis for social research (pp. 113–132). New York, NY: Springer.

24.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

25.

Grasmick

H. G.

Tittle

C. R.

Bursik

R. J. J.

Arneklev

B. J.

(1993). Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. Journal of Research in Crime and Delinquency, 30, 5–29.

26.

Harden

K. P.

Tucker-Drob

E. M.

(2011). Individual differences in the development of sensation seeking and impulsivity during adolescence: Further evidence for a dual systems model. Developmental Psychology, 47, 739–946.

27.

Henson

J. M.

Carey

M. P.

Carey

K. B.

Maisto

S. A.

(2006). Associations among health behaviors and time perspective in young adults: Model testing with boot-strapping replication. Journal of Behavioral Medicine, 29, 127–137.

28.

Hirschi

Gottfredson

M. R.

(1994). The generality of deviance. In Hirschi

Gottfredson

M. R.

(Eds.), The generality of deviance (pp. 1–22). New Brunswick, NJ: Transaction.

29.

Huizinga

Esbensen

Weiher

(1991). Are there multiple paths to delinquency? Journal of Criminal Law and Criminology, 82, 83–118.

30.

Iselin

A. M. R.

Mulvey

E. P.

Loughran

T. A.

Chung

H. L.

Schubert

C. A.

(2012). A longitudinal examination of serious adolescent offenders’ perceptions of changes for success and engagement in behaviors accomplishing goals. Journal of Abnormal Child Psychology, 40, 237–249.

31.

Keough

K. A.

Zimbardo

P. G.

Boyd

J. N.

(1999). Who’s smoking, drinking, and using drugs? Time perspective as a predictor of substance use. Basic and Applied Social Psychology, 21, 149–164.

32.

Kerpelman

J. L.

Eryigit

Stephens

C. J.

(2008). African American adolescents’ future education orientation: Associations with self-efficacy, ethnic identity, and perceived parental support. Journal of Youth and Adolescence, 37, 997–1008.

33.

Knight

K. E.

Ellis

Roark

Henry

K. L.

Huizinga

(2017). Testing the role of aspirations, future expectations, and strain on the development of problem behaviors across young and middle adulthood. Deviant Behavior, 38, 1456–1473.

34.

LaGrange

T. C.

Silverman

R. A.

(1999). Low self-control and opportunity: Testing the general theory of crime as an explanation for gender differences in delinquency. Criminology, 37, 41–72.

35.

Landenberger

N. A.

Lipsey

M. W.

(2005). The positive effects of cognitive-behavioral programs for offenders: A meta-analysis of factors associated with effective treatment. Journal of Experimental Criminology, 1, 451–476.

36.

Laub

J. H.

Sampson

R. J.

(2001). Understanding desistance from crime. Crime and Justice, 28, 1–69.

37.

Longshore

Turner-Rand

Stein

J. A.

(1999). Self-control in a criminal sample: An examination of construct validity. Criminology, 34, 209–228.

38.

Mahler

Simmons

Frick

P. J.

Steinberg

Cauffman

(2017). Aspirations, expectations, and delinquency: The moderating effect of impulse control. Journal of Youth and Adolescence, 47, 1503–1514.

39.

Markus

Nurius

(1986). Possible selves. American Psychologist, 41, 954–969.

40.

Marroquin

Boyle

C. C.

Nolen-Hoeksema

Stanton

A. L.

(2016). Using emotion as information in future-oriented cognition: Individual differences in the context of state negative affect. Personality and Individual Differences, 95, 121–126.

41.

McMurran

(2009). Motivational interviewing with offenders. Legal and Criminological Psychology, 14, 83–100.

42.

Mears

D. P.

Cochran

J. C.

Beaver

K. M.

(2013). Self-control theory and nonlinear effects on offending. Journal of Quantitative Criminology, 29, 447–476.

43.

Menard

Elliott

D. S.

(1996). Prediction of adult success using stepwise logistic regression analysis. Report prepared for the MacArthur Foundation by the MacArthur Chicago-Denver Neighborhood Project.

44.

Miller

R. B.

Brickman

S. J.

(2004). A model of future-oriented motivation and self-regulation. Educational Psychology Review, 16, 9–33.

45.

Moffitt

T. E.

Arseneault

Belsky

Dicksn

Hancox

R. J.

Harrington

… Caspi

(2011). A gradient of self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences, 108, 2693–2698.

46.

Monahan

K. C.

King

K. M.

Shulman

E. P.

Cauffman

Chassin

(2015). The effects of violence exposure on the development of impulse control and future orientation across adolescence and early adulthood: Time-specific and generalized effects in a sample of juvenile offenders. Development and Psychopathology, 27, 1267–1283.

47.

Monahan

K. C.

Steinberg

Cauffman

Mulvey

E. P.

(2009). Trajectories of antisocial behavior and psychosocial maturity from adolescence to young adulthood. Developmental Psychology, 45, 1654–1668.

48.

Mossière

Serin

(2014). A critique of models and measures of treatment readiness in offenders. Aggression and Violent Behavior, 19, 383–389.

49.

Mulvey

E. P.

Schubert

C. A.

Piquero

(2014). Pathways to Desistance—Final technical report. Washington, DC: US Department of Justice.

50.

Muraven

Pogarsky

Schmueli

(2006). Self-control depletion and the general theory of crime. Journal of Quantitative Criminology, 22, 263–277.

51.

Nagin

D. S.

Paternoster

(1991). On the relationship of past to future participation in delinquency. Criminology, 29, 163–189.

52.

Nurmi

J. E.

Poole

M. E.

Kalakoski

(1994). Age differences in future-oriented goals, concerns, and related temporal extension in different sociocultural contexts. Journal of Youth and Adolescence, 23, 471–487.

53.

Olver

M. E.

Stockdale

K. C.

Wormith

J. S.

(2011). A meta-analysis of predictors of offender treatment attrition and its relationship to recidivism. Journal of Consulting and Clinical Psychology, 79, 6–21.

54.

Osgood

D. W.

(2000). Poisson-based regression analysis of aggregate crime rates. Journal of Quantitative Criminology, 16, 21–43.

55.

Pathways to Desistance. (n.d.). Perceptions of chances for success—Subject follow-up. Retrieved from https://www.pathwaysstudy.pitt.edu/codebook/perceptions-of-chances-for-success-sf.html.

56.

Piquero

A. R.

Blumstein

Brame

Haapanen

Mulvey

E. P.

Nagin

D. S.

(2001). Assessing the impact of exposure time and incapacitation on longitudinal trajectories of criminal offending. Journal of Adolescent Research, 16, 54–74.

57.

Pratt

T. C.

Cullen

F. T.

(2000). The empirical status of Gottfredson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38, 931–964.

58.

Pratt

T. C.

Cullen

F. Y.

Sellers

C. S.

Winfree

T. J.

Madensen

Daigle

L. E.

… Gau

(2010). The empirical status of Social Learning Theory: A meta-analysis. Justice Quarterly, 27, 765–802.

59.

Ray

J. V.

Jones

Loughran

T. A.

Jennings

W. G.

(2013). Testing the stability of self-control: Identifying unique developmental patterns and associated risk factors. Criminal Justice and Behavior, 40, 588–607.

60.

Reynolds

J. R.

Pemberton

(2001). Rising college expectations among youth in the United States: A comparison of the 1979 and 1997 NLSY. Journal of Human Resources, 36, 703–726.

61.

Robbins

R. N.

Bryan

(2004). Relationships between future orientation, impulsive sensation seeking, and risk behavior among adjudicated adolescents. Journal of Adolescent Research, 19, 428–445.

62.

Roberton

Daffern

Bucks

R. S.

(2012). Emotion regulation and aggression. Aggression and Violent Behavior, 17, 72–82.

63.

Scheier

M. F.

Carver

C. S.

(1985). Optimism, coping and health: Assessment and implications of generalized outcome expectations. Health Psychology, 4, 219–247.

64.

Schubert

C. A.

Mulvey

E. P.

Steinberg

Cauffman

Losoya

S. H.

Hecker

… Knight

G. P.

(2004). Operational lessons from the Pathways to Desistance project. Youth Violence and Juvenile Justice, 2, 237–255.

65.

Seginer

(2009). Future orientation: Developmental and ecological perspectives. New York, NY: Springer.

66.

Seginer

Noyman

M. S.

(2005). Future orientation, identity, and intimacy: Their relations in emerging adulthood. European Journal of Developmental Psychology, 2, 17–37.

67.

Seginer

Vermulst

Shoyer

(2004). The indirect link between perceived parenting and adolescent future orientation: A multi-step model. International Journal of Behavioral Development, 28, 365–378.

68.

Shulman

E. P.

Harden

K. P.

Chein

J. M.

Steinberg

(2015). Sex differences in the developmental trajectories of impulse control and sensation-seeking from early adolescence to early adulthood. Journal of Youth and Adolescence, 44, 1–17.

69.

Silver

Ulmer

J. T.

(2012). Future selves and self-control motivation. Deviant Behavior, 33, 699–714.

70.

Steinberg

Cauffman

(1996). Maturity of judgment in adolescence: Psychosocial factors in adolescent decision making. Law and Human Behavior, 20, 249–272.

71.

Steinberg

Graham

O’Brien

Woolard

Cauffman

Banich

(2009). Age differences in future orientation and delay discounting. Child Development, 80, 28–44.

72.

Strathman

Gleicher

Boninger

Edwards

C. S.

(1994). The consideration of future consequences: Weighing immediate and distant outcomes of behavior. Journal of Personality and Social Psychology, 66, 742–752.

73.

Sullivan

C. J.

Loughran

(2014). Investigating the functional form of the self-control delinquency relationship in a sample of serious young offenders. Journal of Quantitative Criminology, 30, 709–730.

74.

Sweeten

(2012). Scaling criminal offending. Journal of Quantitative Criminology, 28, 533–557.

75.

Sweeten

Piquero

A. R.

Steinberg

(2013). Age and the explanation of crime, revisited. Journal of Youth and Adolescence, 42, 921–938.

76.

Tangney

J. P.

Baumeister

R. F.

Boone

A. L.

(2004). High self-control predicts good adjustment, less pathology, better grades, and interpersonal success. Journal of Personality, 72, 271–322.

77.

Tittle

C. R.

Ward

D. A.

Grasmick

H. G.

(2004). Capacity for self-control and individuals’ interest in exercising self-control. Journal of Quantitative Criminology, 20, 142–172.

78.

Trommsdorff

Lamm

(1980). Future orientation of institutionalized and noninstitutionalized delinquents and nondelinquents. European Journal of Social Psychology, 10, 247–278.

79.

Turner

M. G.

Piquero

A. R.

(2002). The stability of self-control. Journal of Criminal Justice, 30, 457–471.

80.

Vazsonyi

A. T.

Mikuška

Kelley

E. L.

(2017). It’s time: A meta-analysis on the self-control-deviance link. Journal of Criminal Justice, 48, 48–63.

81.

Ward

Day

Howells

Birgden

(2004). The multifactor offender readiness model. Aggression and Violent Behavior, 9, 645–673.

82.

Weinberger

D. A.

Schwartz

G. E.

(1990). Distress and restraint as superordinate dimensions of self-reported adjustment: A typological perspective. Journal of Personality, 58, 381–417.

83.

Zimbardo

P. G.

(1990). The Stratford time perspective inventory. Stratford, CA: Stratford University.

84.

Zimbardo

P. G.

Boyd

J. N.

(1999). Putting time in perspective: A valid, reliable individual-differences metric. Journal of Personality and Social Psychology, 77, 1271–1288.